A Claude Code skill that analyzes Microsoft Clarity CSV exports and turns raw click/scroll data into actionable CRO insights.
Drop your Clarity click and scroll CSV exports into a conversation and get:
- Structured behavioral analysis — scroll depth curves, click behavior ranked by impact, CSS selectors translated into plain English
- Key findings — 3-5 insights about what visitors are trying to do and where the page fails them
- Prioritized recommendations — actionable next steps tied directly to the data
- Slack-ready summary — a concise version you can paste into your team channel
claude install github:djforge/clarity-analyzerExport click and scroll CSVs from Microsoft Clarity for any page, then:
/clarity-analyze clicks.csv scrolls.csv
Or just mention the files in conversation — the skill will be auto-invoked when it detects Clarity CSV exports.
The skill expects standard Microsoft Clarity CSV exports:
- Clicks CSV — ranked list of clicked elements, click counts, % of total clicks
- Scroll CSV — scroll depth increments (5-100%), visitor counts, % drop-off
Both files include metadata rows (project name, date range, URL filter, page views) that the skill uses automatically.
## /pricing page
Period: Feb 12 – Mar 13, 2026 (30 days)
Page Views: 1,204 | Total Clicks: 892
### Scroll Depth
- 25%: 78% of visitors remain (22% drop-off)
- 50%: 51% remain — scroll cliff between 45-55%
- 100%: 14% reach bottom
### Top Click Behavior
1. Annual/Monthly toggle — 23% of clicks
2. Enterprise "Contact sales" CTA — 11%
3. Pro tier "Start free" — 8%
...
### Key Findings
1. Visitors immediately toggle pricing frequency — they're comparing annual vs monthly
2. Enterprise CTA outperforms self-serve by 1.4x...
MIT